tf_agents.bandits.agents.linear_bandit_agent.LinearBanditVariableCollection

A collection of variables used by LinearBanditAgent.

tf_agents.bandits.agents.linear_bandit_agent.LinearBanditVariableCollection(
    context_dim: int,
    num_models: int,
    use_eigendecomp: bool = False,
    dtype: tf.DType = tf.float32,
    name: Optional[Text] = None
)

context_dim (int) The context dimension of the bandit environment the agent will be used on.
num_models (int) The number of models maintained by the agent. This is either the same as the number of arms, or, if the agent accepts per-arm features, 1.
use_eigendecomp (bool) Whether the agent uses eigen decomposition for maintaining its internal state.
dtype The type of the variables. Should be one of tf.float32 and tf.float64.
name (string) the name of this instance.